fvm::div(T) or fvc::div(T)

sharonyue · October 11, 2013, 22:12

Hi guys,

I wanna make a discretisation on a turbulence stress tensor. $\nabla \cdot \tau$

In this term. $\tau = scale\left( {nut,\nabla U} \right)$

In most of the situation, nut is a scalar. After discretisation, you can get:
$\nabla \cdot \tau = \nabla \cdot \left( {nut\left( {scalar} \right)\nabla U} \right) = fvm::laplacian\left( {nut,U} \right)$

But in my case, nut is a turbulence tensor. So I cant turn it into laplacian term.I have to use fvc::div as follows:
$\nabla \cdot \tau = \nabla \cdot \left( {scale\left( {nut\left( {tensor} \right),\nabla U} \right)} \right) = fvc::div\left( {scale\left( {nut,\nabla U} \right)} \right)$

But actually, I wanna discretize the term into the matrix equation. But I cant use fvm::div(). What should I do.? Thanks.

sharonyue · October 11, 2013, 22:18

BTW.
scaling vector a by vector b would produce vector c whose components are
ci = scale(a, b) = (a1b1, a2b2, a3b3)

sharonyue · October 13, 2013, 19:51

Any ideas?

sharonyue · October 13, 2013, 20:12

Am I clear on this topic?
I mean lets say I have a scalar "a" and a tensor "T". So this formula $\nabla \cdot \left( {a\nabla U} \right)$ in code is:

Code:

fvm::laplacian(a,u);

Regarding this formula: $\nabla \cdot \left( {T} \right)$ in code is:

Code:

fvc::div(U);

But I want it to be Imp. What should I do?

October 11, 2013, 22:12	fvm::div(T) or fvc::div(T)	#1
sharonyue Senior Member Dongyue Li Join Date: Jun 2012 Location: Beijing, China Posts: 839 Rep Power: 17	Hi guys, I wanna make a discretisation on a turbulence stress tensor. $\nabla \cdot \tau$ In this term. $\tau = scale\left( {nut,\nabla U} \right)$ In most of the situation, nut is a scalar. After discretisation, you can get: $\nabla \cdot \tau = \nabla \cdot \left( {nut\left( {scalar} \right)\nabla U} \right) = fvm::laplacian\left( {nut,U} \right)$ But in my case, nut is a turbulence tensor. So I cant turn it into laplacian term.I have to use fvc::div as follows: $\nabla \cdot \tau = \nabla \cdot \left( {scale\left( {nut\left( {tensor} \right),\nabla U} \right)} \right) = fvc::div\left( {scale\left( {nut,\nabla U} \right)} \right)$ But actually, I wanna discretize the term into the matrix equation. But I cant use fvm::div(). What should I do.? Thanks.

October 13, 2013, 20:12		#4
sharonyue Senior Member Dongyue Li Join Date: Jun 2012 Location: Beijing, China Posts: 839 Rep Power: 17	Am I clear on this topic? I mean lets say I have a scalar "a" and a tensor "T". So this formula $\nabla \cdot \left( {a\nabla U} \right)$ in code is: Code: fvm::laplacian(a,u); Regarding this formula: $\nabla \cdot \left( {T} \right)$ in code is: Code: fvc::div(U); But I want it to be Imp. What should I do?

October 11, 2013, 22:18		#2
sharonyue Senior Member Dongyue Li Join Date: Jun 2012 Location: Beijing, China Posts: 839 Rep Power: 17	BTW. scaling vector a by vector b would produce vector c whose components are ci = scale(a, b) = (a1b1, a2b2, a3b3)

October 13, 2013, 19:51		#3
sharonyue Senior Member Dongyue Li Join Date: Jun 2012 Location: Beijing, China Posts: 839 Rep Power: 17	Any ideas?